Study on Language generation and learning from videos

2014 Fiscal Year Final Research Report

• Project/Area Number: 24650084
• Research Category: Grant-in-Aid for Challenging Exploratory Research
• Allocation Type: Multi-year Fund
• Research Field: Perception information processing/Intelligent robotics
• Research Institution: Kobe University
• Principal Investigator: ARIKI Yasuo 神戸大学, 都市安全研究センター, 教授 (10135519)
• Project Period (FY): 2012-04-01 – 2015-03-31
• Keywords: 状況理解 / 画像認識 / 音声認識 / 機械学習 / HMM / 抽出 / 追跡 / メディア変換

Outline of Final Research Achievements

We developed a system which can extract and track a ball and players on videos of soccer sports games captured by four HD cameras. Based on their tracking results, four events such as “goal kick”, “corner kick”, “free kick” and “through-in” can be recognized with high accuracy by Hidden Markov Model (HMM). We carried out the researches of digital camerawork and commentary generation, and then got the prospects for automatic learning of game rules through this study.

Free Research Field

パターン認識